Pattern Detection

Kyew watches for things you explain over and over — the same preferences, the same processes, the same requests — and offers to turn them into reusable knowledge so you never have to repeat yourself.

How It Works

After you've saved a few memories around the same topic, Kyew can analyze them for patterns:

"analyze my patterns around client reporting"

Kyew looks for:

• Similar topics — memories about the same area of your work
• Consistent approaches — things that worked repeatedly
• Related knowledge — information that naturally groups together

Confidence

Kyew scores each pattern by how confident it is:

• 0.9+ — very strong pattern, highly recommended to save
• 0.7-0.9 — solid pattern, worth reviewing
• 0.5-0.7 — possible pattern, might need more data
• Below 0.5 — probably coincidental, not worth saving yet

Using Pattern Analysis

Just ask your AI to look for patterns:

"What patterns do you see in my work memories?"
"Analyze patterns in the client-reporting topic"
"Show me patterns from the last 30 days"

Your AI will tell you what it found and suggest turning strong patterns into skills.

From Pattern to Skill

1. Save memories as you work — "remember that..."
2. Ask for patterns periodically — "analyze my patterns"
3. Review what Kyew found — is this actually something you repeat?
4. Generate a skill from a strong pattern — "save that as a skill"
5. Approve it — your AI uses it automatically going forward

Getting Good Results

For reliable pattern detection:

• Save at least 3 memories on a topic before analyzing
• Use consistent topic names — "client-reporting" every time, not "reports" one day and "client-work" the next
• Be specific in your memories — "Sarah prefers the blue template with Q4 metrics" is much better than "did the report"

When to Analyze

# After a few days of work
"Analyze my patterns from this week"

# After finishing a project
"What patterns emerged in the website-redesign topic?"

# Periodically
"Show me any new patterns you've detected"